Suggestions have been made that in carcinogenesis testing of animals, blind tissue readings would be more valid than readings where the pathologist knows the animals' background and treatments. It is proposed to examine the variation within and between pathologists in reading animal slides with full, partial, and no awareness of background and treatment information in the experiment. Emphasis will be on mammary and liver tumors.